Verification of tracks made on Michael using GEFS mean fields
I made 22 tracks about Hurricane Michael from 19 runs. In three runs, there were two separate non-coexisting tracks, which will be considered a single track for the purpose of this verification. So, I made 19 tracks.
Additionally, within 384
hours of cyclogenesis, there were seven instances, when I missed
cyclogenesis entirely. Originally, formation time was way too early. When I detected the system, I mainly predicted a major hurricane. Movement north from Caribbean, if there was such, was projected as too slow, and movement thereafter had no pattern. The lifespan of the cyclone was mostly predicted to be too long.
As of this analysis, best track positions are available October 4, 18Z thru October 12, 12Z, and as such, position comparisons are possible only for this timespan.
Since I give only SSHS categories in my tracks, I don't calculate intensity errors in units of speed, but in categories. For this purpose, I define tropical or subtropical storms as "Category 0", tropical or subtropical depressions as "Category -1" and non-existence of a tropical or subtropical cyclone as "Category -2".
In the 19 tracks, formation times ranged between September 23, 12Z and October 11, 12Z. Average was October 1, 10Z and median was October 2, 00Z.
In the 26 cases (19 tracks and 7 complete misses), peak intensity ranged between Category -2 and Category 5. Average was Category 1.56 and median was Category 1.
In the 19 tracks, dissipation times ranged between September 25, 18Z and October 25, 00Z. Average was October 8, 17Z and median was October 8, 18Z.
At the lead time of 1 day (24 hours), there were 0 hits, 0 false detections and 1 misses. No available position comparisons.
At the lead time of 2 days (48 hours), there were 2 hits, 0 false detections and 0 misses. 2 available position comparisons produce an average error of 74 nautical miles.
At the lead time of 3 days (72 hours), there were 1 hits, 0 false detections and 2 misses. 1 available position comparison produces an average error of 51 nautical miles.
At the lead time of 4 days (96 hours), there were 2 hits, 3 false detections and 2 misses. 4 available position comparisons produce an average error of 230 nautical miles.
At the lead time of 5 days (120 hours), there were 3 hits, 0 false detections and 2 misses. 4 available position comparisons produce an average error of 286 nautical miles.
At the lead time of 6 days (144 hours), there were 3 hits, 0 false detections and 2 misses. 4 available position comparisons produce an average error of 328 nautical miles.
At the lead time of 7 days (168 hours), there were 3 hits, 1 false detections and 2 misses. 4 available position comparisons produce an average error of 474 nautical miles.
At the lead time of 8 days (192 hours), there were 2 hits, 3 false detections and 2 misses. 3 available position comparisons produce an average error of 593 nautical miles.
At the lead time of 9 days (216 hours), there were 2 hits, 3 false detections and 2 misses. 4 available position comparisons produce an average error of 619 nautical miles.
At the lead time of 10 days (240 hours), there were 1 hits, 12 false detections and 3 misses. 3 available position comparisons produce an average error of 664 nautical miles.
At the lead time of 11 days (264 hours), there were 2 hits, 14 false detections and 2 misses. 3 available position comparisons produce an average error of 729 nautical miles.
At the lead time of 12 days (288 hours), there were 1 hits, 12 false detections and 4 misses. 2 available position comparisons produce an average error of 959 nautical miles.
At the lead time of 13 days (312 hours), there were 1 hits, 11 false detections and 5 misses. 2 available position comparisons produce an average error of 323 nautical miles.
At the lead time of 14 days (336 hours), there were 1 hits, 12 false detections and 6 misses. 2 available position comparisons produce an average error of 322 nautical miles.
At the lead time of 15 days (360 hours), there were 2 hits, 10 false detections and 5 misses. 2 available position comparisons produce an average error of 365 nautical miles.
At the lead time of 16 days (384 hours), there were 1 hits, 10 false detections and 5 misses. 3 available position comparisons produce an average error of 524 nautical miles.
Circle coloration scheme
Here are the tracks once again.
September 13, 18Z run:
September 15, 06Z run (track 2):
September 15, 18Z run:
October 2, 12Z run:
The complete misses were from following runs:
September 23, 06Z
September 23, 12Z
September 24, 06Z
September 25, 06Z
September 25, 12Z
September 27, 06Z
September 28, 06Z
Comments
Post a Comment